/**
* amdgpu_ring_fini - tear down the driver ring struct.
*
* @adev: amdgpu_device pointer
* @ring: amdgpu_ring structure holding ring information
*
* Tear down the driver information for the selected ring (all asics).
*/
void amdgpu_ring_fini(struct amdgpu_ring *ring)
{
int r;
struct amdgpu_bo *ring_obj;
ring_obj = ring->ring_obj;
ring->ready = false;
ring->ring = NULL;
ring->ring_obj = NULL;
amdgpu_wb_free(ring->adev, ring->fence_offs);
amdgpu_wb_free(ring->adev, ring->rptr_offs);
amdgpu_wb_free(ring->adev, ring->wptr_offs);
amdgpu_wb_free(ring->adev, ring->next_rptr_offs);
if (ring_obj) {
r = amdgpu_bo_reserve(ring_obj, false);
if (likely(r == 0)) {
amdgpu_bo_kunmap(ring_obj);
amdgpu_bo_unpin(ring_obj);
amdgpu_bo_unreserve(ring_obj);
}
amdgpu_bo_unref(&ring_obj);
}
}
/**
* amdgpu_ih_ring_fini - tear down the IH state
*
* @adev: amdgpu_device pointer
*
* Tears down the IH state and frees buffer
* used for the IH ring buffer.
*/
void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
{
int r;
if (adev->irq.ih.use_bus_addr) {
if (adev->irq.ih.ring) {
/* add 8 bytes for the rptr/wptr shadows and
* add them to the end of the ring allocation.
*/
pci_free_consistent(adev->pdev, adev->irq.ih.ring_size + 8,
(void *)adev->irq.ih.ring,
adev->irq.ih.rb_dma_addr);
adev->irq.ih.ring = NULL;
}
} else {
if (adev->irq.ih.ring_obj) {
r = amdgpu_bo_reserve(adev->irq.ih.ring_obj, false);
if (likely(r == 0)) {
amdgpu_bo_kunmap(adev->irq.ih.ring_obj);
amdgpu_bo_unpin(adev->irq.ih.ring_obj);
amdgpu_bo_unreserve(adev->irq.ih.ring_obj);
}
amdgpu_bo_unref(&adev->irq.ih.ring_obj);
adev->irq.ih.ring = NULL;
adev->irq.ih.ring_obj = NULL;
}
amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
amdgpu_wb_free(adev, adev->irq.ih.rptr_offs);
}
}
/**
* cik_sdma_ring_test_ring - simple async dma engine test
*
* @ring: amdgpu_ring structure holding ring information
*
* Test the DMA engine by writing using it to write an
* value to memory. (CIK).
* Returns 0 for success, error for failure.
*/
static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
unsigned i;
unsigned index;
int r;
u32 tmp;
u64 gpu_addr;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
r = amdgpu_ring_alloc(ring, 5);
if (r) {
DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
amdgpu_wb_free(adev, index);
return r;
}
amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
amdgpu_ring_write(ring, 1); /* number of DWs to follow */
amdgpu_ring_write(ring, 0xDEADBEEF);
amdgpu_ring_commit(ring);
for (i = 0; i < adev->usec_timeout; i++) {
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < adev->usec_timeout) {
DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
} else {
DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
ring->idx, tmp);
r = -EINVAL;
}
amdgpu_wb_free(adev, index);
return r;
}
/**
* si_dma_ring_test_ib - test an IB on the DMA engine
*
* @ring: amdgpu_ring structure holding ring information
*
* Test a simple IB in the DMA ring (VI).
* Returns 0 on success, error on failure.
*/
static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct dma_fence *f = NULL;
unsigned index;
u32 tmp = 0;
u64 gpu_addr;
long r;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err0;
}
ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1);
ib.ptr[1] = lower_32_bits(gpu_addr);
ib.ptr[2] = upper_32_bits(gpu_addr) & 0xff;
ib.ptr[3] = 0xDEADBEEF;
ib.length_dw = 4;
r = amdgpu_ib_schedule(ring, 1, &ib, NULL, NULL, &f);
if (r)
goto err1;
r = dma_fence_wait_timeout(f, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out\n");
r = -ETIMEDOUT;
goto err1;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
goto err1;
}
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF) {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
} else {
DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
r = -EINVAL;
}
err1:
amdgpu_ib_free(adev, &ib, NULL);
dma_fence_put(f);
err0:
amdgpu_wb_free(adev, index);
return r;
}
/**
* amdgpu_ring_fini - tear down the driver ring struct.
*
* @adev: amdgpu_device pointer
* @ring: amdgpu_ring structure holding ring information
*
* Tear down the driver information for the selected ring (all asics).
*/
void amdgpu_ring_fini(struct amdgpu_ring *ring)
{
ring->ready = false;
amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
amdgpu_wb_free(ring->adev, ring->fence_offs);
amdgpu_wb_free(ring->adev, ring->rptr_offs);
amdgpu_wb_free(ring->adev, ring->wptr_offs);
amdgpu_bo_free_kernel(&ring->ring_obj,
&ring->gpu_addr,
(void **)&ring->ring);
amdgpu_debugfs_ring_fini(ring);
ring->adev->rings[ring->idx] = NULL;
}
/**
* amdgpu_ih_ring_init - initialize the IH state
*
* @adev: amdgpu_device pointer
*
* Initializes the IH state and allocates a buffer
* for the IH ring buffer.
* Returns 0 for success, errors for failure.
*/
int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
bool use_bus_addr)
{
u32 rb_bufsz;
int r;
/* Align ring size */
rb_bufsz = order_base_2(ring_size / 4);
ring_size = (1 << rb_bufsz) * 4;
adev->irq.ih.ring_size = ring_size;
adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
adev->irq.ih.rptr = 0;
adev->irq.ih.use_bus_addr = use_bus_addr;
if (adev->irq.ih.use_bus_addr) {
if (!adev->irq.ih.ring) {
/* add 8 bytes for the rptr/wptr shadows and
* add them to the end of the ring allocation.
*/
adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
adev->irq.ih.ring_size + 8,
&adev->irq.ih.rb_dma_addr);
if (adev->irq.ih.ring == NULL)
return -ENOMEM;
memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
}
return 0;
} else {
r = amdgpu_wb_get(adev, &adev->irq.ih.wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &adev->irq.ih.rptr_offs);
if (r) {
amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
return r;
}
return amdgpu_ih_ring_alloc(adev);
}
}
void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
struct amdgpu_irq_src *irq)
{
amdgpu_wb_free(ring->adev, ring->adev->virt.reg_val_offs);
amdgpu_ring_fini(ring);
}
/**
* cik_sdma_ring_test_ib - test an IB on the DMA engine
*
* @ring: amdgpu_ring structure holding ring information
*
* Test a simple IB in the DMA ring (CIK).
* Returns 0 on success, error on failure.
*/
static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
unsigned i;
unsigned index;
int r;
u32 tmp = 0;
u64 gpu_addr;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(ring, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
goto err0;
}
ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
ib.ptr[1] = lower_32_bits(gpu_addr);
ib.ptr[2] = upper_32_bits(gpu_addr);
ib.ptr[3] = 1;
ib.ptr[4] = 0xDEADBEEF;
ib.length_dw = 5;
r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
AMDGPU_FENCE_OWNER_UNDEFINED,
&f);
if (r)
goto err1;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
goto err1;
}
for (i = 0; i < adev->usec_timeout; i++) {
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < adev->usec_timeout) {
DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
ring->idx, i);
goto err1;
} else {
DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
r = -EINVAL;
}
err1:
fence_put(f);
amdgpu_ib_free(adev, &ib);
err0:
amdgpu_wb_free(adev, index);
return r;
}